Prototype Generation

Prototype generation is a rapidly evolving field focused on creating representative data points (prototypes) for various machine learning tasks. Current research emphasizes improving prototype quality and diversity through techniques like contrastive learning, hyperbolic space embeddings, and attention mechanisms, often integrated into end-to-end training frameworks. These advancements are significantly impacting fields like federated learning, semi-supervised segmentation, and few-shot learning by reducing communication costs, improving data efficiency, and enhancing model robustness across diverse datasets and conditions. The resulting improvements in accuracy and efficiency have broad implications for various applications, including image classification, action recognition, and object detection.

Papers

January 6, 2024

FedTGP: Trainable Global Prototypes with Adaptive-Margin-Enhanced Contrastive Learning for Data and Model Heterogeneity in Federated Learning
Jianqing Zhang, Yang Liu, Yang Hua, Jian Cao
Contrastive Learning Heterogeneous Federated Learning Model Heterogeneity Prototype Generation Global Prototype

October 23, 2023

Generating Prototypes for Contradiction Detection Using Large Language Models and Linguistic Rules
Maren Pielka, Svetlana Schmidt, Rafet Sifa
Linguistic Analysis Syntactic Rule Contradiction Detection Prototype Generation

July 19, 2023

Boundary-Refined Prototype Generation: A General End-to-End Paradigm for Semi-Supervised Semantic Segmentation
Junhao Dong, Zhu Meng, Delong Liu, Jiaxuan Liu, Zhicheng Zhao, Fei Su
End to End Semi Supervised Semantic Segmentation Latent Prototype Prototype Based Prototype Generation

May 18, 2023

HMSN: Hyperbolic Self-Supervised Learning by Clustering with Ideal Prototypes
Aiden Durrant, Georgios Leontidis
Contrastive Learning Self Supervised Representation Learning Visual Representation Learning Hyperbolic Representation Hyperbolic Learning Simulated OSN Prototype Generation

April 13, 2023

ProtoDiv: Prototype-guided Division of Consistent Pseudo-bags for Whole-slide Image Classification
Rui Yang, Pei Liu, Luping Ji
Multiple Instance Learning Whole Slide Image Bag Prototype Prototype Generation Prototype Attention

October 3, 2022

Few-Shot Segmentation via Rich Prototype Generation and Recurrent Prediction Enhancement
Hongsheng Wang, Xiaoqi Zhao, Youwei Pang, Jinqing Qi
Segmentation Based Approach Shot Segmentation Decoder Only LLM Prototype Learning Improving Prediction Prototype Generation

July 12, 2022

Compound Prototype Matching for Few-shot Action Recognition
Yifei Huang, Lijin Yang, Yoichi Sato
Shot Action Recognition Video Search Prototype Generation Global Prototype Prototype Matching

March 16, 2022

PMAL: Open Set Recognition via Robust Prototype Mining
Jing Lu, Yunxu Xu, Hao Li, Zhanzhan Cheng, Yi Niu
Open Set Recognition Prototype Learning Prototype Aggregation Prototype Generation

November 30, 2021

Attentive Prototypes for Source-free Unsupervised Domain Adaptive 3D Object Detection
Deepti Hegde, Vishal M. Patel
Domain Adaptation 3D Object Detection Noisy Pseudo Label State of the Art Object Lidar Based 3D Object Detector Prototype Generation

Prototype Generation

Papers

FedTGP: Trainable Global Prototypes with Adaptive-Margin-Enhanced Contrastive Learning for Data and Model Heterogeneity in Federated Learning

Generating Prototypes for Contradiction Detection Using Large Language Models and Linguistic Rules

Boundary-Refined Prototype Generation: A General End-to-End Paradigm for Semi-Supervised Semantic Segmentation

HMSN: Hyperbolic Self-Supervised Learning by Clustering with Ideal Prototypes

ProtoDiv: Prototype-guided Division of Consistent Pseudo-bags for Whole-slide Image Classification

Few-Shot Segmentation via Rich Prototype Generation and Recurrent Prediction Enhancement

Compound Prototype Matching for Few-shot Action Recognition

PMAL: Open Set Recognition via Robust Prototype Mining

Attentive Prototypes for Source-free Unsupervised Domain Adaptive 3D Object Detection